Use of Data Reconciliation: A Zinc Plant Case Study

Process control, environmental monitoring, budgeting and accounting are based on process data. Without appropriate tools for data analysis, the process data could remain unused and wasted. This paper demonstrates how data reconciliation can be used as a tool to develop closed mass balances. Mass balances are the initial step in plant optimisation (technical, economical or environmental). From mass balances, fundamental process models can be developed. These balances link process parameters (e.g., feed materials) to process performance (e.g., recovery of valuable elements). Some of these models could find their way into simulators useful to plant operators. Classical regression methods regularly fail when they are applied to complex non-linear problems involving a large number of variables. Alternatively, multiparametrical regression analysis can be successfully applied with the aid of neural networks based on reconciled data. Both methods have been applied to plant data originating from the hydrometallurgical zinc plant at Ruhr-Zink, which in this paper serves as a case study. Very promising and useful results were obtained from this study.